Evolutionary Greedy Algorithm for Optimal Sensor Placement Problem in Urban Sewage Surveillance

TL;DR

Addresses cost-efficient sensor placement for sewage surveillance in large urban networks. Introduces Evolutionary Greedy (EG), a hybrid algorithm that combines greedy selection with evolutionary search to optimize a two-objective model: maximize sensing coverage $\sum_i m_i x_i$ and minimize the normalized cost $\sum_i (m_i/\sum_i m_i) \log_2(m_i) x_i$ under a budget, using non-dominated sorting and hypervolume-based evaluation. Demonstrates scalability and effectiveness on synthetic networks and a real Hong Kong network, showing competitive Pareto quality with substantial speedups over a baseline multi-objective greedy, and provides actionable Pareto fronts for policy decisions. The work enables robust, cost-aware urban health surveillance with practical guidelines and data-driven sensor-placement strategies for early outbreak detection.

Abstract

Designing a cost-effective sensor placement plan for sewage surveillance is a crucial task because it allows cost-effective early pandemic outbreak detection as supplementation for individual testing. However, this problem is computationally challenging to solve, especially for massive sewage networks having complicated topologies. In this paper, we formulate this problem as a multi-objective optimization problem to consider the conflicting objectives and put forward a novel evolutionary greedy algorithm (EG) to enable efficient and effective optimization for large-scale directed networks. The proposed model is evaluated on both small-scale synthetic networks and a large-scale, real-world sewage network in Hong Kong. The experiments on small-scale synthetic networks demonstrate a consistent efficiency improvement with reasonable optimization performance and the real-world application shows that our method is effective in generating optimal sensor placement plans to guide policy-making.

Figures (4)

• Figure 1: Illustration of the entry set and $m_i$.
• Figure 2: Flowchart of the proposed Evolutionary Greedy (EG) algorithm.
• Figure 3: Pareto-optimal solutions with linear fit for solutions with sensing coverage ranging from 1229 to 3820.
• Figure 4: Example solution with the maximum sensing coverage. Area A enclosed in a rectangle refers to the area where sensors cluster.